SPFishcool
    • Create new note
    • Create a note from template
      • Sharing URL Link copied
      • /edit
      • View mode
        • Edit mode
        • View mode
        • Book mode
        • Slide mode
        Edit mode View mode Book mode Slide mode
      • Customize slides
      • Note Permission
      • Read
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Write
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Engagement control Commenting, Suggest edit, Emoji Reply
    • Invite by email
      Invitee

      This note has no invitees

    • Publish Note

      Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

      Your note will be visible on your profile and discoverable by anyone.
      Your note is now live.
      This note is visible on your profile and discoverable online.
      Everyone on the web can find and read all notes of this public team.
      See published notes
      Unpublish note
      Please check the box to agree to the Community Guidelines.
      View profile
    • Commenting
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
      • Everyone
    • Suggest edit
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
    • Emoji Reply
    • Enable
    • Versions and GitHub Sync
    • Note settings
    • Note Insights
    • Engagement control
    • Transfer ownership
    • Delete this note
    • Save as template
    • Insert from template
    • Import from
      • Dropbox
      • Google Drive
      • Gist
      • Clipboard
    • Export to
      • Dropbox
      • Google Drive
      • Gist
    • Download
      • Markdown
      • HTML
      • Raw HTML
Menu Note settings Versions and GitHub Sync Note Insights Sharing URL Create Help
Create Create new note Create a note from template
Menu
Options
Engagement control Transfer ownership Delete this note
Import from
Dropbox Google Drive Gist Clipboard
Export to
Dropbox Google Drive Gist
Download
Markdown HTML Raw HTML
Back
Sharing URL Link copied
/edit
View mode
  • Edit mode
  • View mode
  • Book mode
  • Slide mode
Edit mode View mode Book mode Slide mode
Customize slides
Note Permission
Read
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Write
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Engagement control Commenting, Suggest edit, Emoji Reply
  • Invite by email
    Invitee

    This note has no invitees

  • Publish Note

    Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

    Your note will be visible on your profile and discoverable by anyone.
    Your note is now live.
    This note is visible on your profile and discoverable online.
    Everyone on the web can find and read all notes of this public team.
    See published notes
    Unpublish note
    Please check the box to agree to the Community Guidelines.
    View profile
    Engagement control
    Commenting
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    • Everyone
    Suggest edit
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    Emoji Reply
    Enable
    Import from Dropbox Google Drive Gist Clipboard
       owned this note    owned this note      
    Published Linked with GitHub
    Subscribed
    • Any changes
      Be notified of any changes
    • Mention me
      Be notified of mention me
    • Unsubscribe
    Subscribe
    # 2023q1 Homework1 (lab0) contributed by < `SPFishcool` > ## 開發環境 ```shell $ neofetch --stdout OS: Ubuntu 22.04.1 LTS x86_64 Host: GL552VW 1.0 Kernel: 5.15.0-60-generic Uptime: 13 hours, 20 mins Packages: 1877 (dpkg), 9 (snap) Shell: bash 5.1.16 Resolution: 1920x1080 DE: GNOME 42.5 WM: Mutter WM Theme: Adwaita Theme: Yaru [GTK2/3] Icons: Yaru [GTK2/3] Terminal: gnome-terminal CPU: Intel i5-6300HQ (4) @ 3.200GHz GPU: Intel HD Graphics 530 GPU: NVIDIA GeForce GTX 960M Memory: 3626MiB / 11868MiB $ gcc -v Thread model: posix Supported LTO compression algorithms: zlib zstd gcc version 11.3.0 (Ubuntu 11.3.0-1ubuntu1~22.04) ``` ## lab0 資料架構認識 ### list_head - list_head 負責作為鏈結串列將其 entry 串連在一起,其為 Circular Doubly-linked list 類型 :::warning 注意書寫: doubly-linked list,連字號的位置在 doubly 和 linked 之間。 :notes: jserv ::: ```graphviz digraph list { rankdir=LR; node[shape=record, style=bold]; subgraph cluster_0 { head [label="{<prev>prev|<next>next}"]; color=invis; label=list_head } } ![](https://i.imgur.com/8XRDf8u.png) ``` - list_head架構圖 :::warning 使用 Graphviz 製圖並嵌入到 HackMD 頁面。 :notes: jserv 收到!正在改進中! ::: - list_head的struct為: ```c struct list_head { struct list_head *prev; struct list_head *next; }; ``` ### element_t - `element_t` 為佇列儲存 value,內部有 `list_head` 架構的 list 能夠將所有 `element_t` 串接起來 ```graphviz digraph list { rankdir=LR; node[shape=record, style=bold]; subgraph cluster_0 { value [label="{*value}"]; list [label="list", shape=plaintext] head [label="{<prev>prev|<next>next}"]; label=element_t; } } ``` - element_t 架構圖 - element_t 的 struct 為: ```c typedef struct { char *value; struct list_head list; } element_t; ``` - 因此 lab0 實做的佇列結構會是 ![](https://i.imgur.com/2NomHmV.png) ### queue_chain_t - 在執行 [qtest.c](https://github.com/SPFishcool/lab0-c/blob/master/qtest.c) 時,會先 create 一個 `queue_chain_t` 結構的物件,其功用類似上述佇列結構的 **head**,目的是每當 `q_new` 一個新的佇列時會將新的佇列利用 list_add_tail 加入到 `queue_chain_t` 鏈結串列的最後面。 >![](https://i.imgur.com/96FnpoC.png) >queue_chain_t 架構圖,size 為佇列的數量 ```c typedef struct { struct list_head head; int size; } queue_chain_t; ``` ### queue_contex_t - `queue_contex_t` 在紀錄每一條佇列的資訊,`*q` 為 pointer of queue head,`size` 是佇列裡面 element_t 的數量,`chain` 為與其他 `queue_contex_t` 串連的 `list_head`。 >![](https://i.imgur.com/80aWxMq.png) >queue_contex_t架構圖 ```c typedef struct { struct list_head *q; struct list_head chain; int size; int id; } queue_contex_t; ``` - 再加入 `queue_chain_t` 其完整佇列架構就完成了! >![](https://i.imgur.com/NuEJyGr.png) >queue 為上圖的紅色虛線部份 ## lab0 開發過程 ### q_new `q_new()` 會先配置 `list_head` 大小的記憶體位置,接下來判斷是否配置成功,最後 `INIT_LIST_HEAD(new)` 會將這個 node 接回自己(成為 Doubly-linked list 型態)。 ![](https://i.imgur.com/Qvdzk6d.png) ```c struct list_head *q_new() { struct list_head *new = malloc(sizeof(struct list_head)); // check if space allowed if (!new) return NULL; INIT_LIST_HEAD(new); return new; } ``` ### q_free `q_free()` 從 `head->next` (第一個有entry的節點)開始釋放,直到回到 head 最後釋放 head 節點。 ```c void q_free(struct list_head *l) { if (!l) return; struct list_head *indirect = l->next; struct list_head *tmp = NULL; while (indirect != l) { tmp = indirect; indirect = indirect->next; q_release_element(list_entry(tmp, element_t, list)); } free(l); } ``` ### q_insert_head / q_insert_tail 兩者可以使用 [list.h](https://github.com/SPFishcool/lab0-c/blob/master/list.h) 中的 `list_add` 與 `list_add_tail` 將 `element_t` 內的list加入鏈結串列中。 ```c bool q_insert_head(struct list_head *head, char *s) { if (!(head && s)) return false; element_t *new_ele = malloc(sizeof(element_t)); if (!new_ele) return false; new_ele->value = (char *) malloc(strlen(s) * sizeof(char) + 1); if (!new_ele->value) { free(new_ele); return false; } strncpy(new_ele->value, s, strlen(s) + 1); // struct list_head *new_node = (new_ele -> list); list_add(&new_ele->list, head); return true; } ``` ### q_remove_head/q_remove_tail remove 只有將 element_t 的串列節點從佇列中移除並沒有釋放其空間,因此只需要將其 next 與 prev 連接起來即可移除。 ```c element_t *q_remove_head(struct list_head *head, char *sp, size_t bufsize) { if (!head || list_empty(head)) return NULL; struct list_head *indirect = head->next; //or head->prev (q_remove_tail) list_del(indirect); element_t *ele = list_entry(indirect, element_t, list); if (sp) { strncpy(sp, ele->value, bufsize-1); *(sp+bufsize-1) = '\0'; } return ele; } ``` ### q_size `q_size` 則從 `head->next` 開始計算這條佇列的 `element_t` 數量,並且回傳 `i` (count 後的結果) ```c int q_size(struct list_head *head) { if (!head) return -1; struct list_head *indirect = head->next; int i = 0; while (indirect != head) { i++; indirect = indirect->next; } return i; } ``` ### q_delete_mid 因為佇列是 doubly-linked list 結構,因此可以使用左右指標(最左和最右可以快速找到),其原理是每一迴圈 `left_node` 往右走一步,`right_node` 往左走一步,直到兩指標相遇(奇數 element)或相鄰(偶數 element)時 right_node 就是我們要找的節點。 優點:跟快慢指標相比,迴圈節省了一半的次數(僅限 Circular Doubly-linked list)。 ![](https://i.imgur.com/m7wqbwr.png) ```c bool q_delete_mid(struct list_head *head) { while(!head || list_empty(head)) return false; struct list_head *left_node = head->next; struct list_head *right_node = head->prev; while ((left_node->next != right_node) && (left_node != right_node)) { left_node = left_node->next; right_node = right_node->prev; } list_del(right_node); q_release_element(list_entry(right_node, element_t, list)); return true; } ``` ### q_delete_dup `q_delete_dup` 這個 question 我紀錄此次和前次的比較紀錄,而受到「判決」的節點為 `indirect->prev`(目前指到的節點的前一個節點) - `cmp`:`indirect` 與 `indirect->prev` 的比較結果 - `prev_cmp`:`indirect->prev` 與 `indirect->prev->prev` 的比較結果 ![](https://i.imgur.com/5lbM3Lc.png) 不管是 `cmp` 還是 `prev_cmp` 只要為 0 都代表著**中間這個節點的 value 與前後節點重複**,必刪除! ```c bool q_delete_dup(struct list_head *head) { if (!head || list_empty(head)) return false; struct list_head *indirect = (head->next)->next; element_t *prev_ele = NULL; element_t *current_ele = NULL; int cmp = -1, prev_cmp = -1; while (indirect != head) { prev_ele = list_entry(indirect->prev, element_t, list); current_ele = list_entry(indirect, element_t, list); cmp = strcmp(prev_ele->value, current_ele->value); if (cmp == 0 || prev_cmp == 0) { list_del(indirect->prev); q_release_element(prev_ele); } prev_cmp = cmp; indirect = indirect->next; } if (cmp == 0) { list_del(indirect->prev); q_release_element(current_ele); } return true; } ``` ### q_swap `q_swap` 只要將佇列內的 element 兩兩交換即可,因此只需要使用 [list.h](https://github.com/SPFishcool/lab0-c/blob/master/list.h) 中的 `list_move` 把自己搬到 `next` 後面即可 ```c void q_swap(struct list_head *head) { if (!head || list_empty(head)) return; struct list_head *indirect = head->next; while (indirect != head && indirect->next != head) { list_move(indirect, indirect->next); indirect = indirect->next; } } ``` ### q_reverse 這裡則是使用 pointer to pointer 紀錄第一個 element 節點內的 next 位址(`&head->next->next`),在每次迴圈中將 `next` 指到的節點利用 `list_move` 移到 `head` 前面直到 `next` 指向的位置為 `head` 為止。 ```c void q_reverse(struct list_head *head) { if (!head || list_empty(head)) return; struct list_head **indirect = &((head->next)->next); while (*indirect != head) list_move(*indirect, head); } ``` ### q_reverseK `q_reverseK` 其原理與 `q_reverse` 相似,但因為如果後面個數不足 k 個則不做 reverse,因此我的作法是先讓一個指標邊走邊數,數到第 k 個後,再將後面的節點一一搬到前面來,因此如果走完一圈沒有數到 k 個,剩下的節點就不會動到。 ```c void q_reverseK(struct list_head *head, int k) { if (!head || list_empty(head)) return; struct list_head **indirect = &(head->next); struct list_head *current = head->next; struct list_head *tmp = NULL; int i = 1; while (current != head) { if (i % k == 0) { tmp = current; while (*indirect != tmp) { list_move(tmp->prev, current); current = current->next; } indirect = &((current)->next); } current = current->next; i++; } } ``` ### q_sort 參考[你所不知道的 C 語言: linked list 和非連續記憶體](https://hackmd.io/@sysprog/c-linked-list#Merge-Sort-%E7%9A%84%E5%AF%A6%E4%BD%9C)merge_sort寫法。 原先想法是要維持 doubly-linked list 的結構去設計,但當我要 `malloc` 新的 head 來接分割後的子串列時遇到`ERROR:Calls to malloc disallowed`,因此決定將 circular 斷開將最後一個 node 的`prev`指向`NULL`才進行`mergesort`並且在排序後將排序過後的串列重新接上 head。 ```c void q_sort(struct list_head *head) { if (!head || list_empty(head)) return; struct list_head *start = head->next; struct list_head *tail = head->prev; tail->next = NULL; struct list_head *sorted_list = merge_sort(start); head->next = sorted_list; struct list_head *cur = head; while (cur->next) { cur->next->prev = cur; cur = cur->next; } cur->next = head; head->prev = cur; } ``` ```c struct list_head *mergeList(struct list_head *l1, struct list_head *l2) { struct list_head *head = NULL; struct list_head **current = &head; while (l1 && l2) { if (strcmp(list_entry(l1, element_t, list)->value, list_entry(l2, element_t, list)->value) >= 0) { *current = l2; l2 = l2->next; } else { *current = l1; l1 = l1->next; } current = &((*current)->next); } *current = (struct list_head *) ((__intptr_t) l1 | (__intptr_t) l2); return head; } ``` ```c struct list_head *merge_sort(struct list_head *head) { if (!head || head->next == NULL) return head; struct list_head *slow = head; for (struct list_head *fast = head->next; fast && fast->next; fast = fast->next->next) { slow = slow->next; } struct list_head *list2 = slow->next; slow->next = NULL; return mergeList(merge_sort(head), merge_sort(list2)); } ``` ### q_descend `q_descend` 若是 `prev` 方向有比自己小的節點則這些小的節點都要 delete 掉,直到遇到下一個比自己大的節點,則移動到下一個節點(`prev`) ```c int q_descend(struct list_head *head) { if (!head || list_empty(head)) return 0; struct list_head *current = head->prev; struct list_head *cur_prev = current->prev; while (cur_prev != head) { if (strcmp(list_entry(current, element_t, list)->value, list_entry(cur_prev, element_t, list)->value) > 0) { list_del(cur_prev); q_release_element(list_entry(cur_prev, element_t, list)); } else { current = current->prev; } cur_prev = current->prev; } int len = q_size(head); return len; } ``` ### q_merge 在 [qtest.c](https://github.com/SPFishcool/lab0-c/blob/master/qtest.c) 裡執行`q_merge`其程式碼為 ```c len = q_merge(&chain.head); ``` 這個 chain 其實就是 `queue_chain_t` 結構,所以得知輸入的資料為 `queue_chain_t` 內的 `head` 元素因此我們可以很簡單的拜訪所有佇列再使用[list.h](https://github.com/SPFishcool/lab0-c/blob/master/list.h)中的 `list_splice_init` 從第二個佇列到最後一個一一串在 ~~`id=0`~~ 目前第一個 `queue_contex_t->q` 的串列內。 ```c int q_merge(struct list_head *head) { if(!head || list_empty(head)) return 0; else if(list_is_singular(head)) return list_entry(head->next, queue_contex_t, chain)->size; queue_contex_t *target = list_entry(head->next, queue_contex_t, chain); queue_contex_t *que = NULL; list_for_each_entry (que, head, chain) { if (que == target) continue; list_splice_init(que->q, target->q); target->size = target->size + que->size; que->size = 0; } q_sort(target->q); return target->size; } ``` ## 目前 queue.c 靜態與動態評分 ### make test - score: 95/100 - problem: q_insert 系列程式時間複雜度非 constant time ### make valgrind - score: 94/100 - problem: Test of malloc failure on new 測試未通過 - reason:在 `qtest` 執行結束未妥善釋放其程式所配置的記憶體所導致 - solved: 在研讀[Valgrind + 自動測試程式](https://hackmd.io/@sysprog/linux2023-lab0/%2F%40sysprog%2Flinux2023-lab0-b#%E4%BB%A5-Valgrind-%E5%88%86%E6%9E%90%E8%A8%98%E6%86%B6%E9%AB%94%E5%95%8F%E9%A1%8C)後發現大部分的test其實都在記憶體配置上出現問題,例如: ```bash ==104151== 128,000 bytes in 2,000 blocks are still reachable in loss record 4 of 6 ==104151== at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so) ==104151== by 0x10F110: test_malloc (harness.c:133) ==104151== by 0x10F645: q_insert_tail (queue.c:72) ==104151== by 0x10C62B: do_it (qtest.c:312) ==104151== by 0x10DDFA: interpret_cmda (console.c:181) ==104151== by 0x10E3AF: interpret_cmd (console.c:201) ==104151== by 0x10E7B0: cmd_select (console.c:610) ==104151== by 0x10F09C: run_console (console.c:705) ==104151== by 0x10D1EC: main (qtest.c:1223) ``` 而上述 problem 的錯誤為: ```bash ==104070== 160 bytes in 5 blocks are possibly lost in loss record 3 of 3 ==104070== at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so) ==104070== by 0x10CBCE: do_new (qtest.c:145) ==104070== by 0x10DDFA: interpret_cmda (console.c:181) ==104070== by 0x10E3AF: interpret_cmd (console.c:201) ==104070== by 0x10E7B0: cmd_select (console.c:610) ==104070== by 0x10F09C: run_console (console.c:705) ==104070== by 0x10D1EC: main (qtest.c:1223) ``` 在這情況下我決定先嘗試解決比較輕的問題,解決過程如下: - 檢查 queue.c 是否有哪一個功能沒有完全 free 空間 - 檢查 script 資料夾內的 cmd 檔案,發現沒有此問題的腳本都有執行 free。 - 開始檢查 [qtest.c](https://github.com/SPFishcool/lab0-c/blob/master/qtest.c) 的退出程式發現: ```c static bool q_quit(int argc, char *argv[]) { return true; report(3, "Freeing queue"); if (current && current->size > BIG_LIST_SIZE) set_cautious_mode(false); if (exception_setup(true)) { struct list_head *cur = chain.head.next; while (chain.size > 0) { queue_contex_t *qctx, *tmp; tmp = qctx = list_entry(cur, queue_contex_t, chain); cur = cur->next; q_free(qctx->q); free(tmp); chain.size--; } } ``` 發現 `do_quit` 第一行程式碼即 `return true` 表示下方程式完全沒有走過,註解掉後重新 `make valgrind` 所有問題都解決了。 ## 研讀 Linux 核心原始程式碼的 lib/list_sort.c 在 linux 核心原始程式碼內的 [lib/list_sort.c](https://github.com/torvalds/linux/blob/master/lib/list_sort.c) 所作的 sort 為 **Merge Sort** ,但這裡面的 Merge Sort 跟我們以前認識的 Merge Sort 不同。 在 `list_sort` 內有 3 個 Function - `merge` - `merge_final` - `list_sort` 而走訪 linked list 時會使用 count 會計算目前走訪到的節點數量,一旦走訪數量走到 $2^k$ 數量時不執行 `merge`,反之就會執行 merge。其流程如下 1. 一開始 `list` 從 `head->next` 開始,而 `pending` 起始為 `NULL`,`count` 起始為 `0`,linked list 最後指向 `NULL`。 2. 迴圈開始,宣告 `**tail`指向 `pending` 的位址。 3. 會先計算 `count` 是否為 $2^k-1$ ,如果是,跳到 "7."。 4. 準備 `merge` ,將 `a = *tail`、`b = a->prev`,執行 `merge(priv, cmp, b, a);` 5. `merge` 會有一個 `**tail` 紀錄前一次比較最小的節點的 `next` 位址,每一次比較會把 `*tail` 指向較小的節點,**其實就是一般的merge,這裡就把 `next` 拿來做 merge 的主要 link。** 6. merge return 為 merge 後的第一個節點(最小)存入 `a` ,將目前的 `a->prev` 重指成 `b->prev`,藉由 `*tail` 把 `pending` 指向 `a`(merge 起始節點)。 7. 將 `list->prev` 指向 `pending`,`list` 往前一步, `panding = list->prev`, 把 `pending->next` 指向 `NULL`。 8. 若 `list` 還沒走完,回到 "2." 9. merge 剩下還沒 merge 起來的 pending list。 10. 最後 `merge_final` 把 merge 完的 list 將其 `prev` 連接起來。 看完 [lib/list_sort.c](https://github.com/torvalds/linux/blob/master/lib/list_sort.c) 發現他並沒有特別做 divide 而是走訪過程一一將 `next` 斷開並且把 `next` 當成執行 merge 動作時走訪的 link,比較「[你所不知道的 C 語言: linked list 和非連續記憶體](https://hackmd.io/@sysprog/c-linked-list#Merge-Sort-%E7%9A%84%E5%AF%A6%E4%BD%9C)」所提及的遞迴與非遞迴,此做法省去 divide 這一部份,也沒有使用 stack 有 Max Depth 的限制。 ## 引入 lib/list_sort.c 到 lab0-c 專案 - 在 [qtest.c](https://github.com/SPFishcool/lab0-c/blob/master/queue.c) 新增 [lib/list_sort.c](https://github.com/torvalds/linux/blob/master/lib/list_sort.c) 三條 finction 和引入 [list_sort.h](https://github.com/torvalds/linux/blob/master/include/linux/list_sort.h) 內宣告設定 - 參考[lab 0:Valgrind + 自動測試程式](https://hackmd.io/@sysprog/linux2023-lab0/%2F%40sysprog%2Flinux2023-lab0-b)中的 qtest 直譯器實作,使用 `ADD_COMMAND` 新增剛才加入的 `list_sort`, 再依照檔案裡的 `do_sort` 新增 `do_list_sort` ```c ADD_COMMAND(list_sort, "Sort queue in ascending order by list_sort", ""); ``` - 製作分析準備 ### 利用 perf 分析 `list_sort` 與自己做的 `q_sort` 效能比較 參考 [linhoward0522 開發紀錄 (lab0)](https://hackmd.io/@linhoward/lab0-2023#2023q1-Homework1-lab0) 使用 perf 分析其效能,在 [traces/trace-15-perf.cmd](https://github.com/SPFishcool/lab0-c/blob/master/traces/trace-15-perf.cmd) 有三筆資料來測試排序,因此我們可以依照這個檔案自製用來測試的 cmd 檔。 ``` . ├── RAND-1000000-list.cmd ├── RAND-1000000-q.cmd ├── RAND-100000-list.cmd ├── RAND-100000-q.cmd ├── RAND-10000-list.cmd ├── RAND-10000-q.cmd ├── RAND-1000-list.cmd └── RAND-1000-q.cmd ``` `q` : `q_sort` `list` : `list_sort` 接下來是試做 `shell scripts` ```bash #!/bin/bash # q_sort for i in ./traces/perf_traces/RAND-*-q.cmd do perf stat --repeat 5 -o "${i%.cmd}"-report -e cache-misses,branches,instructions,context-switches ./qtest -v 0 -f "$i" done #list_sort for i in ./traces/perf_traces/RAND-*-list.cmd do perf stat --repeat 5 -o "${i%.cmd}"-report -e cache-misses,branches,instructions,context-switches ./qtest -v 0 -f "$i" done ``` - `q_sort` | #node | cache-misses | branches | instructions | context-switch | | ------- | ------------ | ----------- | ------------ | -------------- | | 1000 | 3469 | 77,9884 | 516,0401 | 0 | | 10000 | 3,3403 | 621,9374 | 4305,8950 | 0 | | 100000 | 543,0532 | 6326,2119 | 4,3295,3471 | 0 | | 1000000 | 7491,5547 | 4,6974,1917 | 31,6153,3055 | 10 | - `list_sort` | #node | cache-misses | branches | instructions | context-switch | |-------|--------------|----------|--------------|-----------------| |1000|2489|77,7798|519,9240|0| |10000|4,1257|614,2028|4329,3845|0| |100000|439,1922|6258,0696|4,3642,0292|3| |1000000|5824,6891|4,6028,3507|31,6579,3448|9| 可以看到節點數增多,效能差異越大,雖然 RAND 會導致偶爾會讓 `list_sort` 效能看起來跟 `q_sort` 差不多甚至更差,但時間上還是 `list_sort` 略勝一籌。 ## 在 [qtest.c](https://github.com/SPFishcool/lab0-c/blob/master/qtest.c) 新增 `shuffle` 命令 作業要求為使用「Fisher–Yates shuffle」完成 `shuffle`,其原理就是從最後一個節點開始,隨機選取前面的節點與其交換,交換完後再往前一格完成前面敘述的事情直到第二張牌為止。 ```c for (int i = len-1;i > 0;i--) { //random 1~i int count = rand() % i; struct list_head *node = head->next; for (int j=1; j < count;j++) node = node->next; swap(tail, node); tail = node->prev; } ``` 跟設定 `list_sort` 一樣,新增 `do_shuffle` 與 `ADD_COMMAND` ```c ADD_COMMAND(shuffle, "shuffle queue.", ""); ``` 接下來統計剛設計出來的 `shuffle` 亂度 參考 [L01: lab0 亂數+論文閱讀](https://hackmd.io/@sysprog/linux2023-lab0/%2F%40sysprog%2Flinux2023-lab0-d) 中的測試程式並載入 [Makefile](https://github.com/SPFishcool/lab0-c/blob/master/Makefile) 執行指令 ```makefile shuffle_check: qtest scripts/shuffle_test.py scripts/shuffle_test.py ``` 執行結果: ![](https://i.imgur.com/6nFK4rS.png) 依照 4 個數字的亂數來看,分佈差距大約在 $\pm1.2\%$ ,學生突然想到如果數字在多一點的話分佈會不會變化比較大,因此我將數字加到 6 個,而 6 個數字的排列組合比 4 個數字多出 30 倍,因此也將次數增加到 30000000 次。 執行結果: ![](https://i.imgur.com/MgJ8VzX.png) 在這張圖最少次為 41081,最多次為 42281,幅度有稍微增加,但沒有特別明顯。 ## 使用 entropy 觀察亂數 在認識 entropy 之前,對 entropy 的唯一認知就是機器學習中計算 loss 值的方式 `cross entropy`,發現其中的含意與 entropy 也有一些關係,但知道了「entropy 與亂數產生的資訊量成反比」還是覺得對 entropy 有點抽象,因此我打算從 qtest.c 裡的指令 `option entropy 1` 與 `ih RAND 10` 找出關於 entropy 的蛛絲馬跡,最後發現 [shannon_entropy.c](https://github.com/SPFishcool/lab0-c/blob/master/shannon_entropy.c) 裡的 `shannon_entropy` 實作了以下 entropy 公式。 $-\sum^{n}_{i=1} p(x_i)\log_2{p(x_i)}$ ```c for (uint32_t i = 0; i < BUCKET_SIZE; i++) { if (bucket[i]) { uint64_t p = bucket[i]; p *= LOG2_ARG_SHIFT / count; entropy_sum += -p * log2_lshift16(p); } } entropy_sum /= LOG2_ARG_SHIFT; return entropy_sum * 100.0 / entropy_max; ``` 這裡除了使用 bucket 計算每一個字元的熵值,而且使用 `LOG_ARG_SHIFT` 預計算的常數來避免使用浮點數增加計算時間。

    Import from clipboard

    Paste your markdown or webpage here...

    Advanced permission required

    Your current role can only read. Ask the system administrator to acquire write and comment permission.

    This team is disabled

    Sorry, this team is disabled. You can't edit this note.

    This note is locked

    Sorry, only owner can edit this note.

    Reach the limit

    Sorry, you've reached the max length this note can be.
    Please reduce the content or divide it to more notes, thank you!

    Import from Gist

    Import from Snippet

    or

    Export to Snippet

    Are you sure?

    Do you really want to delete this note?
    All users will lose their connection.

    Create a note from template

    Create a note from template

    Oops...
    This template has been removed or transferred.
    Upgrade
    All
    • All
    • Team
    No template.

    Create a template

    Upgrade

    Delete template

    Do you really want to delete this template?
    Turn this template into a regular note and keep its content, versions, and comments.

    This page need refresh

    You have an incompatible client version.
    Refresh to update.
    New version available!
    See releases notes here
    Refresh to enjoy new features.
    Your user state has changed.
    Refresh to load new user state.

    Sign in

    Forgot password

    or

    By clicking below, you agree to our terms of service.

    Sign in via Facebook Sign in via Twitter Sign in via GitHub Sign in via Dropbox Sign in with Wallet
    Wallet ( )
    Connect another wallet

    New to HackMD? Sign up

    Help

    • English
    • 中文
    • Français
    • Deutsch
    • 日本語
    • Español
    • Català
    • Ελληνικά
    • Português
    • italiano
    • Türkçe
    • Русский
    • Nederlands
    • hrvatski jezik
    • język polski
    • Українська
    • हिन्दी
    • svenska
    • Esperanto
    • dansk

    Documents

    Help & Tutorial

    How to use Book mode

    Slide Example

    API Docs

    Edit in VSCode

    Install browser extension

    Contacts

    Feedback

    Discord

    Send us email

    Resources

    Releases

    Pricing

    Blog

    Policy

    Terms

    Privacy

    Cheatsheet

    Syntax Example Reference
    # Header Header 基本排版
    - Unordered List
    • Unordered List
    1. Ordered List
    1. Ordered List
    - [ ] Todo List
    • Todo List
    > Blockquote
    Blockquote
    **Bold font** Bold font
    *Italics font* Italics font
    ~~Strikethrough~~ Strikethrough
    19^th^ 19th
    H~2~O H2O
    ++Inserted text++ Inserted text
    ==Marked text== Marked text
    [link text](https:// "title") Link
    ![image alt](https:// "title") Image
    `Code` Code 在筆記中貼入程式碼
    ```javascript
    var i = 0;
    ```
    var i = 0;
    :smile: :smile: Emoji list
    {%youtube youtube_id %} Externals
    $L^aT_eX$ LaTeX
    :::info
    This is a alert area.
    :::

    This is a alert area.

    Versions and GitHub Sync
    Get Full History Access

    • Edit version name
    • Delete

    revision author avatar     named on  

    More Less

    Note content is identical to the latest version.
    Compare
      Choose a version
      No search result
      Version not found
    Sign in to link this note to GitHub
    Learn more
    This note is not linked with GitHub
     

    Feedback

    Submission failed, please try again

    Thanks for your support.

    On a scale of 0-10, how likely is it that you would recommend HackMD to your friends, family or business associates?

    Please give us some advice and help us improve HackMD.

     

    Thanks for your feedback

    Remove version name

    Do you want to remove this version name and description?

    Transfer ownership

    Transfer to
      Warning: is a public team. If you transfer note to this team, everyone on the web can find and read this note.

        Link with GitHub

        Please authorize HackMD on GitHub
        • Please sign in to GitHub and install the HackMD app on your GitHub repo.
        • HackMD links with GitHub through a GitHub App. You can choose which repo to install our App.
        Learn more  Sign in to GitHub

        Push the note to GitHub Push to GitHub Pull a file from GitHub

          Authorize again
         

        Choose which file to push to

        Select repo
        Refresh Authorize more repos
        Select branch
        Select file
        Select branch
        Choose version(s) to push
        • Save a new version and push
        • Choose from existing versions
        Include title and tags
        Available push count

        Pull from GitHub

         
        File from GitHub
        File from HackMD

        GitHub Link Settings

        File linked

        Linked by
        File path
        Last synced branch
        Available push count

        Danger Zone

        Unlink
        You will no longer receive notification when GitHub file changes after unlink.

        Syncing

        Push failed

        Push successfully